analytics and the red meat value chain groups...red meat industry • unsophisticated traders •...
TRANSCRIPT
ANALYTICS
AND
THE RED MEAT VALUE CHAIN
Dr David Rutley October 2017
Introduction
• Red Meat Industry
• Thomas Foods International
• Production, Prices, Volume & Variation
• Adoption of Analytics
• SAS or Fools
David Rutley
• Lamb Supply Chain Coordinator
• 30 years Red Meat Industry production & research
• Consumer Appeal, Retail, Food Service, Distribution, Processing, Finishing,
Production, Genetics
• Thomas Foods – Quantify and Connect the Supply Chain
Signals from the Consumer to the Producer
Red Meat Industry
• Unsophisticated traders
• Buyers and sellers negotiate the spot price
• Accountants check profit daily to monitor & control prices& production
• Little to no prediction
• Lots of use of Excel, so I report in Excel
Red Meat Industry
• Largest Ag industry in Australia
Including flow on effects
• $23 Billion p.a.
• Employs > 134,000 people
• $8.7 Billion household income
Thomas Foods International
• Australia’s largest family owned red meat processor
• > $ 1.5 Billion p.a.
• 3 Plants Murray Bridge, Lobethal, Tamworth
• Domestic distributor, Holco
• USA distributor, TFI USA
Thomas Foods International
• ~25,000 sheep, lambs & goats & 1,100 cattle per day
• ~3.5 Million lambs p.a. 200,000 cattle p.a.
~1.5 Million sheep p.a.
• 15 - 20% of Australia’s sheep – largest processor
80% Lamb exported approx 8 - 10% global trade
Inside the Machine
Last 3 years
• ~10,000 different suppliers
• 25 different types of beef cattle
– Organic, Angus, grass fed, grain fed, short, long & very long fed …
• New season’s lamb, lamb, hogget, mutton, goat
– Organic, grass fed, grain fed …
Inside the Machine
• 1 lamb 25 kg carcass 20 kg saleable ~1.3 cartons
• 1 beef 280 kg carcass 200 kg saleable ~13 cartons
Deconstructing a black box
– Genotype by Environment = Variation
CV HSCW = 15%
• >3,200 products
• 70+ countries
95% PL 70 to 130%
2014 Jul – 2016 Dec
• >3,200 different Products, 285,000 tonnes, 14.9 M cartons, $1.6 B
Show SAS Table of Carton Weights
Product Sales – Murray Bridge
Max Price$44.81/kg
Min Price$0.67/kg
26% varianceover 8 months
}
Supply Costs – Beef & Lamb – HIGHLY Volatile
Somewhat Complex
• With this level of complexity and variation –
How do we know we are going to make a profit?
• Continual monitoring and controlling
Little to no prediction
• Trial and Error & Experience
Somewhat Complex
How do we know we are going to make a profit?
• Food is so important that it’s production
MUST be viable over the long term.
• TFI’s beef processing probably
did not make a profit at any stage
in 2016 and for at least ½ of 2017
The Future = Analytics
• Profit = Income – Expense
• Value = Profit – Risk
• Risk = Chance of a Loss Stochastic Model
MUST understand RISK & therefore VARIATION
MUST develop Predictive Models
Beef – Predict Processing Yield
• 1 Beef Carcass
• > 50 possible pieces of meat
• > 230 possible cuts of meat, bone, trim, fat or waste
TFI Beef Yield Prediction Model based on
• Hot Standard Carcass Weight
• P8 Fat Depth
• Muscle Score
• Eye Muscle Area
• Days in Chiller
Beef – Predict Processing Yield
• Collecting data to relate sale price to cattle buying price
• 25 beef programs, 10 important programs
Price by Product Code
from previous Table
Lamb Value Calculator – Predict Saleable Yield
• 1 Lamb Carcass
• > 36 possible pieces of meat
• > 92 possible cuts of meat, bone, trim, fat or waste
TFI Lamb Value Calculator based on
• Hot Standard Carcass Weight
• GR Fat Depth
• Days in Chiller
• Frequency of 4,536 different possible lamb carcasses
Lamb Value Calculator – Predict Saleable Yield
Not 0
FatnessDifferences
Supplier Benchmarking – Rangers Valley Cattle Station
• Compare >2000 suppliers
• Value $20 per 400kg steer
• $750,000 per year
• Eliminated bottom 20% suppliers
Beef Eating Quality
• > 17 traits
• Have 1 minute to record 8 of these in the marshalling room.
Lamb Eating Quality
• 6 lambs/minute
• Cannot measure pH or Intra Muscular Fat (marbling) this fast
• Looking at hyperspectral imaging without cutting carcasses
– 530 different wavelengths
Beef & Lamb Yield & Processing
DEXA
• Predict Lean Meat Yield
• Drive robots for automated processing
David Rutley – Biologist & Applied Statistician
Since 1984
• GLIMM, REG, Genstat, ASREML, S, SPSS, R & SAS(since 1994)
I have found SAS to be the most capable of
• handling Large & Complex data sets
• Robust, Repeatable & Verifiable analyses
PROPHESIS Province of Deaf Mutes & Fools
Provided we understand Confidence & Prediction Limits
Good Prediction Better than Guessing
INVESTING vs GAMBLING
KNOWLEDGE
THANK YOU……….